Comparative Effectiveness of 2 Diabetes Prevention Lifestyle Programs in the Workplace

The City and County of San Francisco Diabetes Prevention Trial

Assiamira Ferrara, MD, PhD; Julia C. McDonald, MS; Susan D. Brown, PhD; Janet G. Alexander, MS; Jennifer L. Christian-Herman, PhD; Stephanie Fisher, MPH; Charles P. Quesenberry, PhD


Prev Chronic Dis. 2020;17(5):E38 

In This Article


This study was approved by the Kaiser Permanente Northern California institutional review board. All participants provided written informed consent.

Study Design

The CCSF Diabetes Prevention Trial was a 2-arm, parallel, randomized controlled trial to compare the effectiveness of the YMCA-DPP and VLM-DPP in 7 worksites of the CCSF. Study staff enrolled CCSF employees from each worksite location in waves. At each worksite, enrollment lasted approximately 1 month. Participants were enrolled in September 2015 at the first worksite and in July 2016 at the seventh worksite. Follow-up ended in August 2017.

Participants were CCSF employees. Department contacts from the 7 participating locations promoted the study using flyers, department-wide emails, announcements at meetings, and information on the CCSF website. Interested CCSF employees attended an onsite open information session where research staff explained the 2 DPP interventions; described what participation in the research study would entail; conducted eligibility screening; and assessed height, weight, and waist circumference. Employees were considered eligible based on results from a CDC prediabetes screener,[13] which defines high risk as a score of 9 or higher and having a body mass index (BMI, determined by assessing weight in kg divided by height in m2) of ≥25.0 kg/m2 or ≥23 kg/m2 if Asian. Criteria for exclusion were having a major health condition (ie, self-reported diabetes, heart disease, renal insufficiency, cancer, or high blood pressure), not having regular access to a computer or the internet, current enrollment in another weight loss program, or plans to move or change departments in the next 12 months.

Eligible employees were asked to attend an orientation session 1 week later, during which research staff further explained the study. Employees who were interested in participating provided written informed consent and completed surveys.

The study project manager randomly assigned employees to either the in-person YMCA-DPP delivered at the worksite or the online VLM-DPP. An adaptive randomization procedure[14] was used to ensure that the 2 groups remained balanced overall and in terms of the following key characteristics: age (<50 y or ≥50 y), BMI (<30 kg/m2 or ≥30 kg/m2), sex (male or female), race/ethnicity (white, African American, Hispanic, Asian, or multiracial/other), and job type (office or not office [ie, public works, transportation, or other field-based job]).

DPP-derived Lifestyle Interventions

All participants were given the goal of losing 5%–7% of their body weight through dietary change and participation in 150 minutes of moderate-intensity physical activity per week.

The YMCA-DPP, an in-person intervention of 16 weekly, 1-hour group sessions (core curriculum) and 3 biweekly and 5 monthly group sessions (maintenance curriculum),[11] was held for groups of 8 to 15 employees at each participating CCSF worksite at a time convenient for employees (ie, during breaks or lunch). Group sessions were facilitated by trained YMCA-DPP lifestyle coaches and covered the key topics of healthy eating, increasing physical activity, reducing stress, problem solving, and motivation to maintain behavior changes. Participants weighed in with the lifestyle coach at the beginning of each hour-long session and were asked to self-report their minutes of physical activity. They were also asked to self-monitor their food intake and activity daily, beginning from session 1 of the program.

Participants assigned to the VLM-DPP, an online program offered through Canary Health,[12] first received a single in-person group orientation where they received instructions on how to use the online platform. The VLM-DPP curriculum included 16 weekly educational sessions (core curriculum) and 8 monthly maintenance sessions to be completed online at a convenient time and place for the participant. Sessions lasted from 25 to 45 minutes and included streaming audio and interactive visual material. Trained VLM-DPP coaches engaged participants using secure messaging to discuss progress, encourage regular participation, and respond to participants' comments and questions. The program included behavior change techniques such as goal setting and self-monitoring of diet, physical activity, and weight using the online platform.


Assessments were conducted by trained research staff blinded to intervention assignment at baseline, 6 months, and 12 months. Data on demographic characteristics and information on whether a participant's job was office-based or not office-based (such as public works, transportation, or other field-based jobs) were collected at baseline. Height was measured at baseline using a Seca stadiometer (Seca GmbH). Weight and waist circumference were measured at baseline and at 6 months and 12 months. Weight was measured using a Tanita scale (Tanita Corporation). All body measurements were taken according to standard protocols[15] with participants in light clothing and without shoes, with measurements taken in duplicate to the nearest 0.1 pound or 0.1 centimeter. The average of the 2 measurements was used if the difference between them was less than 1.0 pound or 1.0 centimeter; otherwise, a third measurement was taken and used. Weight and height measured at baseline were used to calculate BMI. Waist circumference was measured by positioning a tape measure 1 inch above the umbilicus at the end of the participant's normal expiration. The mean value of waist circumference was calculated if the 2 initial measurements agreed within 1 centimeter. Otherwise, an additional measurement was taken, and the third recording was used.

Diet and physical activity were assessed at baseline and the 6-month and 12-month follow up. The Block Fat Screener was used to assess diet,[16] and the International Physical Activity Questionnaire[17] was used to assess physical activity. The number of intervention sessions completed between random assignment and 6 and 12 months was assessed by YMCA group facilitators and the VLM online platform.


The primary outcome was body weight expressed as change in the amount of body weight in kilograms and percentage change in body weight. Secondary outcomes included changes in waist circumference; physical activity, expressed as duration and intensity per week; dietary fat intake; and participant engagement, expressed as mean number of intervention sessions completed and the proportion of participants who completed 4 or more sessions.

Power and Statistical Analyses

The study sample of 80 individuals per intervention condition provided 80% power to detect as small as a 1.8% difference in mean percentage weight loss between intervention conditions, assuming a standard deviation for percentage weight loss of 4%.[9,10] In addition, with 80 individuals in each condition, the minimum detectable difference in mean weight change was 2.2 kg, assuming a standard deviation for weight change of 5 kg.[9,10]

Linear mixed effects regression techniques were used to model change in weight, percentage change in body weight, waist circumference, physical activity, and dietary fat intake, as a function of intervention condition and time, controlling for baseline variables used in the adaptive randomization procedure (age, BMI, sex, race/ethnicity, and job type).[18] Additional models included an intervention condition-by-time interaction term, which enabled us to assess the variation in group differences in means over time. Model parameters were estimated via maximum likelihood methods. Modified Poisson regression[19] for estimation of relative risk (RR; ratio of proportions) was used to compare conditions on the proportion attending 4 or more core sessions, and linear regression analysis was used to compare conditions on the mean number of completed core sessions. Two sets of analyses were conducted for weight and waist circumference outcomes: 1) an analysis including only individuals with clinic-measured weights and waist circumferences, and 2) an analysis including all randomized individuals by using measured or imputed weights and waist circumferences and if measurements were missing, using multiple imputation techniques. The chained equation technique[20] was used to generate 20 imputed data sets, with discriminant function and multiple linear regression analysis used as imputation models for missing categorical and continuous covariates, respectively. Statistical analyses were performed on each of the imputed data sets, with results combined using Rubin's rules[21] providing valid point and interval estimates appropriately accounting for the uncertainty in imputing the missing data.[21] The analyses for diet and physical activity outcomes included only individuals who completed the related questionnaires. All analyses were conducted using SAS version 9.3 (SAS Institute, Inc).